This action might not be possible to undo. Are you sure you want to continue?

BooksAudiobooksComicsSheet Music### Categories

### Categories

### Categories

Editors' Picks Books

Hand-picked favorites from

our editors

our editors

Editors' Picks Audiobooks

Hand-picked favorites from

our editors

our editors

Editors' Picks Comics

Hand-picked favorites from

our editors

our editors

Editors' Picks Sheet Music

Hand-picked favorites from

our editors

our editors

Top Books

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Audiobooks

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Comics

What's trending, bestsellers,

award-winners & more

award-winners & more

Top Sheet Music

What's trending, bestsellers,

award-winners & more

award-winners & more

Welcome to Scribd! Start your free trial and access books, documents and more.Find out more

Structural Engineering, Mechanics and Materials Professor: S. Govindjee

Virtual Work1 1 Principle of Virtual Work

When using the Principle of Stationary Potential Energy to ﬁnd the equilibrium states for a mechanical system we are forced to make the assumption that the system involved is conservative. This restricts the type of problems that can be analyzed. In this section we will present a special case of the Principle of Virtual Work. The Principle of Virtual Work is a more general way of stating equilibrium and is not restricted to the conservative setting. The special case we will present is also known as the Method of Virtual Displacements and we will use these terms interchangeably.

1.1

Motivating example: Tension-compression bar

AE

b(x)

F

u( )

x

L

x

Trial solutions

Figure 1: Tension-compression bar and examples of trial solutions Consider the tension-compression bar shown in Figure 1. The bar is subjected to an end-force F and a distributed load b(x); the loads may or may not be conservative. The governing equilibrium equation for the problem is:

1

Based on the original prepared by Dr. T. Koyama, Spring 2009.

1

Diﬀerential equation of equilibrium (strong form): dR +b=0 dx The boundary conditions for the problem tell us that the solution u(x) must satisfy the condition u(0) = 0. Let us deﬁne the set (space) of admissible displacements S as the set of all functions which satisfy the boundary conditions. The true displacement solution must be in this set of functions.

• Set (space) of admissible displacements, S = {u(x) | u(0) = 0} Every functions u ∈ S satisﬁes the displacement (kinematic) boundary conditions. Let us also deﬁne a second set of functions called the set (space) of test functions (virtual displacements) V.

• Set (space) of test solutions (virtual displacements), V = {δu(x) | δu = 0 at all points where u is known.} The requirements on the functions δu ∈ V turn out to be a very convenient choice. To proceed further, we will now convert the strong form of the equilibrium equation to the so-called weak form. The weak form of the equilibrium equation can be derived from the strong form strictly by algebraic and calculus manipulations – i.e. without the use of any physical information about the problem. 1. Start with the diﬀerential equation of equilibrium dR +b=0 . dx 2. Multiply the diﬀerential equation by an arbitrary function δu ∈ V δu dR + δu b = 0 . dx 2

Note that δu(0) = 0 (since δu ∈ V) and that R(L) = F (according to the force boundary condition) L δu(L)F − 0 dδu R dx + dx L δu b dx = 0 .. Integrate out the exact diﬀerential L [δuR]. Integrate the over the length of the bar (i. Exploit the product rule of diﬀerentiation L 0 dδu d [δuR] − R dx + dx dx d dx L δu b dx = 0 . or the virtual work equation: Weak statement of equilibrium (weak form): L L δεR dx = 0 0 δu b dx + δu(L)F . L δuR|L − 0 0 dδu R dx + dx δub dx = 0 0 L ⇒ δu(L)R(L) − δu(0)R(0) − 0 dδu R dx + dx L δu b dx = 0 . [0. 0 5. L]) L δu 0 dR dx + dx L δu b dx = 0 . 0 6.e. This last equation is known as the weak form.3. stating exactly the same condition: 3 . 0 4. This implies that the diﬀerential equation of equilibrium and weak statement of equilibrium are equivalent. 0 7. the weak statement of equilibrium. Remarks: 1. Group terms and deﬁne δε := L dδu dx L δεR dx = 0 0 δu b dx + δu(L)F . Note that we can follow the steps outlined above in the reverse order (going from Step 7 to Step 1).

we have derived this result. Thus in this context it is sometimes also known as the virtual work theorem. Example 1: Weak form content. such that they imply one and the same thing. 2. As presented here. δux a b L x δ"x a b L x bx F Ra bx L Rb Figure 2: Speciﬁc choice of a test function δu and the resulting equilibrium condition 4 . The Principle of Virtual Work states that a system is in equilibrium if and only if the weak equilibrium statement holds ∀ δu ∈ V.Diﬀerential equation of equilibrium (strong form) ⇔ Weak statement of equilibrium (weak form) Thus the weak statement of equilibrium is a restatement of the diﬀerential equation of equilibrium.

The weak statement of equilibrium also known as the virtual work equation can thus be stated as: 5 . b] of the bar. The quantity δε = (d/dx)δu is often called the virtual strain in analogy to real strains ε = (d/dx)u. 2. Clearly this function satisﬁes the conditions required for δu to be a member of V. Thus. The quantity L δu b dx + δu(L)F 0 is called the external virtual work. our chosen test function tests global force equilibrium over the segment [a. Observe that if we assume δu has dimensions of length. 2. In general. b]. The diﬀerence between the bar force R(a) and R(b) is equal to the integral of the distributed load b over [a. This function and its derivative (δε = (d/dx)δu) δε = δ(x − a) − δ(x − b) are depicted in Fig.To partially illustrate the equivalence of the weak and strong forms let us select a test function δu ∈ V of the form δu(x) = H(x − a) − H(x − b) . b]. 3. The quantity L L δε R dx = 0 0 A δε σ dA dx is known as the internal virtual work. Remarks: 1. 5. Substituting this into the weak form yields L L [δ(x − a) − δ(x − b)] R(x) dx = 0 b 0 [H(x − a) − H(x − b)] b dx + F δu(L) ⇒ R(a) − R(b) = a b dx . The space of admissible displacements S is sometimes called the space of trial solutions. In general the external virtual work is equal to the product of the virtual displacements times the real forces (integrated if given as distributions). which is just a statement of equilibrium for the segment [a. then this term has dimensions of work. 2. the internal virtual work is the integral over the body of the virtual strains times the real stresses. see Fig. 4.

Applying the Principle of Virtual Work. Specify the governing diﬀerential equation of equilibrium. Every choice of a test function in the virtual work equation yields a generalized equilibrium equation for the system.s ⇔ Weak statement of equilibrium 8. Multiply the diﬀerential equation by a test function δu ∈ V and integrate over the whole domain. 2. The displacement (kinematic) boundary conditions are also called the essential boundary conditions.C. Thus one has the following more precise equivalence: Diﬀerential equation of equilibrium ⊕ Force B. The force boundary conditions are sometimes referred to as the natural boundary conditions.2 Summary of the steps leading to the virtual work equations In order to derive a weak statement of equilibrium (in the context of virtual displacements). 10. 9. with our particular form of the virtual work equation. one has to execute the following basic steps: 1. is often called the method of virtual displacements. 1.3 Solutions The Principle of Virtual Work (Method of Virtual Displacements) can be used as a tool for ﬁnding exact as well as approximate solutions to the equations of equilibrium. Determine the trial solution space S and test function space V for the given problem. 1. The methodology is summarized as: Find u ∈ S such that L L δε R dx = 0 0 δu b dx + F δu(L) for all δu ∈ V where δε = δu′ and R = R(u′ ).Internal Virtual Work = External Virtual Work 6. 4. 3. Integration by parts (product rule of diﬀerentiation followed by integration of an exact diﬀerential) to obtain the weak statement of equilibrium.). 7. 6 . because they are “naturally” pre-embedded in the weak statement of equilibrium. note that the force boundary conditions were incorporated (Step 6. In deriving the weak statement of equilibrium.

bx F L ux 5 δux 8 Figure 3: The search problem for a tension compression bar. the method becomes simpler and more tractable. The steps are quite similar to the method of Ritz in combination with the Principle of Stationary Potential Energy. Pick a function u ∈ S. then go back to Step 1 and repeat the process. R(u′ ) = AEu′ . This is the case if one desires to ﬁnd the “exact” solution. This can be considered an inﬁnite dimensional search problem where one iterates through the following steps: 1. i. 3. possible trial solutions u and test functions δu For the case of a tension-compression bar ﬁxed at one end. If the equation does not hold for some δu. 2.For the special case of a linear elastic material.e. one has L L δu′ AEu′ dx = 0 0 δu b dx + F δu(L). 3. Test this function against all possible δu ∈ V.. If one desires only an approximate solution. make sure the virtual work equation holds true for all δu ∈ V. in that one selects an approximate form for the functions 7 . some possible trial solutions and test functions are shown in Fig.

The virtual work equation for this problem is: L L δεR dx = 0 0 δu b0 dx . This is called a Bubnov-Galerkin approximation. Example 2: Bar with a constant distributed load. δu(x) ≈ δCx. Note that we have chosen the functions in S and V to have the same form. 0 ⇒ δC 0 R dx − 8 . one selects. an approximation for the trial solution space S and test function space V. b(x)=b 0 L u u exact (quadratic) approx x Figure 4: Tension-compression bar with distributed load. let us consider an example. 4. The exact solution for this problem is a quadratic function. the exact solution and approximate solution obtained from a 1 parameter approximation Consider the tension-compression bar with one end ﬁxed as shown in Fig. into the virtual work equation yields: L L (δC)R dx = 0 L 0 L (δCx)b0 dx xb0 dx = 0 .involved and then solves a ﬁnite dimensional problem. The bar is subjected to a constant distributed load. V ≈ {δu(x) | δu(x) = δCx} . To make this more concrete. Let us select the following one parameter approximations for S and V. Inserting the selected approximate forms for u and δu. S ≈ {u(x) | u(x) = Cx} . In the case of the Principle of Virtual Work. u(x) ≈ Cx.

Since this last statement must hold for any δC (the Principle of Virtual Work). Assuming a linear elastic material R(u′ ) = AEu′ . V = {δu(x) | δu(0) = 0} . 2AE The key to a good approximation using the Principle of Virtual Work lies in selecting good approximations for the solution space S and test function space V. ˆ S ⊂ S. 2. one has L L AEC dx − 0 0 xb0 dx = 0 . It should be emphasized that should one have treated this problem with the Principle of Stationary Potential Energy assuming an approximation u = Cx. one would have obtained the same equation as above for C. Solving for C gives: C= Thus the approximate solution is. uapprox (x) = Remarks: 1.) b0 L x. Figure 4 compares the approximate solution with the exact solution. (This of course is only true when one assumes an elastic material. 9 . Observe how our selection in the example. 2AE b0 L2 2 AEL = b0 L . we have S = {u(x) | u(0) = 0} . The approximation is seen to be reasonable for a single parameter approximation. one must have: L L R dx − 0 0 xb0 dx = 0 . which represents a single scalar equation in one unknown. For the example problem. ˆ One must select the approximate trial solution space S and approximate test function space ˆ V so that they are subspaces of S and V. ˆ ˆ S = V = {f (x) | Cx} . ˆ V⊂V .

Chosen to satisfy kinematic BCs . To extend the method to obtain better approximations. S and V can be extended to the form: Nu ˆ S = ˆ V = u(x) | u = δu(x) | δu = un fn (x). 10 . one can consider an approximation constructed from a linear combination ˆ ˆ of many functions. Here K is a Nδu -by-Nu matrix whose mn-th entry is deﬁned by L Kmn = 0 ′ ′ gm (x)AEfn (x) dx . just as in the Method of Ritz. Chosen to be zero where the kinematic BCs are speciﬁed m=1 . For the linear elastic material case. and one has: L Nδu ′ δum gm (x)AE 0 Nδu m=1 L n=1 Nu ′ gm (x)AE 0 n=1 Hm Nδu ′ un fn (x) dx − 0 Nu ′ un fn (x) L Nδu dx = 0 b0 m=1 L δum gm (x) dx ⇒ m=1 δum b0 gm (x) dx = 0 ⇒ m=1 δum Hm . F is a Nδu -by-1 vector whose m-th entry is deﬁned by L Fm = 0 ′ gm (x) b0 dx . n=1 Nδu δum gm (x). This last equation must hold ∀δum which implies that Hm = 0 for all m.satisﬁed this property. Inserting these forms into the virtual work equation for the example gives: L Nδu ′ δum gm (x)R 0 m=1 n=1 Nu ′ un fn (x) L Nδu dx = 0 b0 m=1 δum gm (x) dx. where we have used the notation of un and δum for the parameters. R = AEu′ . Thus Nu 0 L ′ ′ gm (x)AEfn (x) dx un = n=1 Nu 0 L b0 gm (x) dx ⇒ n=1 Kmn un = Fm ⇒ Ku = F .

One can show that they are formally equivalent for conservative systems. Remarks: 1. 2. This is the (Bubnov)-Galerkin method. Assume that u∗ (x) is the function(point) which gives the minimum value of Π.4 Equivalence between the Principle of Virtual Work and the Principle of Stationary Potential Energy Through the previous examples. 11 . the value of Π should increase: ⇒ Π(u∗ ) ≤ Π(u∗ + αδu) . The choices for fn and gm are essentially arbitrary after the basic conditions have been satisﬁed (viz.and u is a Nu -by-1 vector whose n-th entry is un . S = {u | u(0) = 0} . If gm = fm and Nu = Nδu (the usual choice). Here let us derive their equivalence for the simple case of an elastic tension-compression bar subject to a conservative constant distributed load. The total potential energy for this system is L Π(u) = 0 1 AE(u′ )2 dx − 2 L b0 u dx . one may have observed a notable similarity in the Principle of Virtual Work and the Principle of Stationary Potential Energy. Then if one adds any multiple α ∈ R of any function δu to u∗ . This is not merely a coincidence. 0 Let us search for the stationary point of Π over. Here δu is selected such that δu ∈ V = {δu | δu(0) = 0} . the ﬁnal expression for u(x) satisﬁes the essential boundary conditions and δu(x) is equal to zero at these points). b(x)=b 0 L Figure 5: Tension-compression bar with distributed load Consider the tension-compression bar in Figure 5.. then K is square and symmetric. 1.

thus. In the last line we have made use of the linear elastic relation R = AEu∗ ′ . we can rewrite this as L L Rδu′ dx = 0 0 b0 δu dx . By construction. Remarks: 12 . h(α) has a minimum (stationary) point at α = 0.Deﬁne now h(α) := Π(u∗ + αδu) L 1 AE (u∗ + αδu)′ = 2 0 L 2 dx − 0 b0 (u∗ + αδu) dx . Note that this is exactly the statement of the Principle of Virtual Work. The derivative can be evaluated as: D(α) Π(K*) minimum at α=0 α Figure 6: The h(α) function and its minimum (stationary) value at α = 0 dh dα L = α=0 0 L 1 AE2 u∗ ′ + αδu′ δu′ dx − 2 L L b0 δu dx 0 α=0 = 0 L AE(u∗ )′ δu′ dx − 0 L b0 δu dx = 0 Rδu′ dx − 0 b0 δu dx . α=0 This situation is depicted in Fig. 6. which must hold ∀δu ∈ V. dh dα =0. Noting that the derivative is zero by construction.

δΠ is called the variation (or variational derivative) of Π. α=0 δΠ = 0 Rδu′ dx − 0 b0 δu dx . δu) . 2.1. an applied distributed torque. replace f by Π. and an end-torque as shown in Fig. and v by δu. 3. one has. x by u∗ . α=0 This quantity is also denoted Df (x)[v]. the directional derivative of f is deﬁned as: d f (x + αv) dα . δΠ := For the given example problem. Given a scalar-valued function f (x) which depends on the position x ∈ Rn in n-dimensions. The variational derivative is equivalent to the directional derivative introduced in vector calculus. It should be emphasized that the method used to show the equivalence is modestly general and can be applied to other mechanical systems as well. The kinematic boundary condition for the problem is ϕ(0) = ϕ0 ¯ and the force boundary condition is ¯ T (L) = T . L L dh dα . 13 . It is deﬁned as. Consider a bar where the left end has an imposed rotation. Thus we have the equivalence Df (x)[v] ∽ δΠ(u∗ )[δu] = δΠ(u∗ . 7. 2 Torsion t(z) '(0)=' 0 T L Figure 7: Torsion bar with distributed torque and end torque In this section we treat the problem of torsion of a bar. In the case of the potential energy Π.

from distributed torque External V. one obtains the virtual work equation for this system: L L δαT (α) dz = 0 Internal V. from point torque at z = L The problem statement for ﬁnding the solution ϕ to the problem becomes: 14 . L]. dz The kinematic boundary conditions of the problem deﬁne the form for the trial solution space S = {ϕ(z) | ϕ(0) = ϕ0 } ¯ and the test function space V = {δϕ(z) | δϕ(0) = 0} .The diﬀerential equation governing equilibrium (strong form) is given by dT +t=0 .W. To derive the virtual work equation (weak statement of equilibrium). L δϕ 0 L dT + δϕt dz = 0 dz ⇒ 0 (δϕT )′ − δϕ′ T + δϕt dz = 0 L L ⇒ [δϕT ]L − 0 0 L δϕ′ T dz + 0 L δϕt dz = 0 δϕt dz 0 L ⇒ 0 L δϕ′ T dz = [δϕT ]L + 0 ⇒ 0 L δϕ′ T dz = δϕ(L)T (L) − δϕ(0)T (0) + 0 L δϕt dz ⇒ 0 δϕ′ T dz = 0 ¯ δϕt dz + δϕ(L)T . one takes the strong form of equilibrium and multiplies it by a test function δϕ ∈ V and integrates over the whole domain [0. 0 ¯ δϕt dz + δϕ(L)T External V. By deﬁning the virtual twist rate δα := δϕ′ and twist rate α := ϕ′ .W.W.

for all δϕ ∈ V where δα = δϕ′ and T = T (α).Find ϕ ∈ S such that. one has L L δϕ′ GJϕ′ dz = 0 0 ¯ δϕ t dz + T δϕ(L). L z t(z)=t L 0 '(0)=' 0 '(L)='L L Figure 8: Torsion bar with rotation specifed at both ends Example 4: No kinematic boundary conditions. T (α) = GJα = GJϕ′ . T (0) = T0 and T (L) = TL . see Fig. 9. Example 3: Fixed end rotations. ¯ ¯ V = {δϕ(z) | δϕ(0) = 0 and δϕ(L) = 0} . State the virtual work equation along with the required function spaces. From the given boundary conditions. 8. ϕ(0) = ϕ0 and ϕ(L) = ϕL . 15 . L L δαT dz = 0 0 ¯ δϕ t dz + T δϕ(L). State the virtual work equation along with the required function spaces. ¯ ¯ Consider a torsion rod where torques are imposed at both ends. we have that S and V are: S = {ϕ(z) | ϕ(0) = ϕ0 and ϕ(L) = ϕL } . Thus. Consider a torsion rod where the rotation is imposed at both ends. In this case the only contribution to the external virtual work is from the distributed load. see ¯ ¯ Fig. the expression for the weak statement of equilibrium (the virtual work equation) is L L δαT dz = 0 0 δϕ t0 z dz . For the special case of a linear elastic material. It should be emphasized that S and V change with changing kinematic boundary conditions.

Selections similar to what we did with the Principle of Stationary Potential Energy are appropriate here and for generalized programs one typically uses functions with compact support. Thus one sees that the given choice of test function implies global equilibrium of the bar. see Fig. we have that S and V are S = {ϕ(z) | no conditions} . Thus. Then δα = 0. the Principle of Virtual Work can be used as a tool for computing approximate solutions. As with the tension-compression bar. In this case there is no contribution to the external virtual work from a distributed load but there are contribution from the two applied loads. One simply selects a set of ﬁnitely parameterized trial solutions and test functions. 0 T t(z)=0 0 TL L Figure 9: Torsion bar with torque specifed at both ends Remarks: 1. ¯ ¯ 0 = δC TL − δC T0 ¯ ¯ ⇒ δC TL − T0 = 0 ¯ ¯ ⇒ TL − T0 = 0 ¯ ¯ ⇒ TL = T 0 . V = {δϕ(z) | no conditions} . ∀δC ∀δC Remarks: 1. which implies. the expression virtual work equation reads: L ¯ ¯ δαT dz = δϕ(L)TL − δϕ(0)T0 . As an example of how the virtual work equation embeds diﬀerent concepts of equilibrium.From the given boundary conditions. 10. These are then substituted into the virtual work equation which provides the necessary relations for determining the parameters in the approximation. 16 . consider the test function δϕ = δC.

5 5 ^ 8 8 ^ Can be parameterized by finite set ˆ Figure 10: Trial solution space S and test function space V approximated by subspaces S ˆ and V parameterized by a ﬁnite set of parameters. qx V M x L Figure 11: Cantilever beam with distributed load and end loads 17 .

and an end-moment as is shown in Fig. Substituting these observations into the general virtual work expression yields for the internal virtual work: L σδε = V V A −yδκσ dAdx = 0 δκM dx 18 .3 Bending The use of virtual work to express the equilibrium equations for beam bending follow the same pattern that we have developed for tension-compression bars and torsion rods. The diﬀerential equations governing equilibrium are: dM +V = 0 . In particular. The virtual normal strains are also restricted by the kinematic assumptions. V where σ is the bending stress. 11. which tells us that ε = −yκ = −yv ′′ . Consider a beam ﬁxed at one end and subjected to a distributed load. The internal virtual work is in general the real stresses times the virtual strains integrated over the volume. The kinematic boundary conditions for this problem are: v(0) = 0 . For a beam in bending this gives σδε + τ δγ dV . let us reason out its form. dx dV +q = 0 . θ(L) = v ′ (L) = 0 . Before formally deriving the virtual work expression. and the force boundary conditions are: ¯ V (L) = −EIv ′′′ (L) = V . and δε and δγ the corresponding virtual normal and shear strains. this restricts the virtual normal strains to the form δε = −yδκ = −yδv ′′ . M (L) = EIv ′′ (L) = 0 . the kinematic assumptions require the shear strains to be zero and thus we also have δγ = 0. dx The kinematic boundary conditions for the problem deﬁne the form for the trial solution space S = {v(x) | v(0) = 0 and v ′ (0) = 0} and the test function space V = {δv(x) | δv(0) = 0 and δv ′ (0) = 0} . In Bernoulli-Euler beams. an end-shear. τ the shear stress.

W. External V. Multiplying each by the corresponding motion where it acts gives for the external virtual work: L ¯ ¯ δvq dx + δv(L)V + δθ(L)M . from distributed load External V. 0 where we have deﬁned δθ = δv ′ .The external virtual work is the product of the real external load with the virtual motions.W. dx2 Now multiply by a test function δv ∈ V and integrate over the whole domain [0. from point moment at x = L To formally derive the weak statement of equilibrium from the governing diﬀerential equations. 0 ¯ ¯ δv q dx + δv(L)V + δθ(L)M . In our case there is a distributed load and two point loads. L]. The ﬁnal result is then: L L δκM (κ) dx = 0 Internal V. 19 . let us ﬁrst combine the two equations deﬁning equilibrium into one by eliminating the expression for the shear force. from point force at x = L External V. This yields: d2 M −q =0 . L δvM ′′ − δvq dx = 0 0 L ⇒ 0 L [(δvM ′ )′ − δv ′ M ′ ] − δvq dx = 0 [(δvM ′ )′ − {(δv ′ M )′ − δv ′′ M }] − δvq dx = 0 0 L ⇒ ⇒ 0 L (−δvV )′ − (δv ′ M )′ + δv ′′ M − δvq dx = 0 L L L ⇒ 0 L δv ′′ M dx = 0 L δvq dx + 0 (δv ′ M )′ dx + 0 L (δvV )′ dx ⇒ 0 L δv ′′ M dx = 0 L δvq dx + [δv ′ M ]0 + [δvV ]L 0 δvq dx + v ′ (L)M (L) − v ′ (0)M (0) + δv(L)V (L) − δv(0)V (0) 0 L ⇒ 0 L δv ′′ M dx = δv ′′ M dx = 0 0 ⇒ ¯ ¯ δvq dx + δv ′ (L)M + δv(L)V .W.W.

For the special case of a linear elastic material. Deﬁne the appropriate function spaces and use a subset of them to ﬁnd an approximate solution to the given problem. One observes that this is identical to the form which we reasoned out above. Remarks: 1. the Principle of Virtual Work can be used as a tool for computing exact as well as approximate solutions to beam problems.Using the notation for rotation and curvature one alternatively has L L δκM dx = 0 0 ¯ ¯ δvq dx + δθ(L)M + δv(L)V . 12. The problem statement for ﬁnding the solution v(x) to the problem becomes: Find v ∈ S such that. Example 5: Indeterminate beam with end-moment. which is simply supported at one end and ﬁxed at the ¯ other. for all δv ∈ V where δθ = δv ′ . M x L Figure 12: Beam with end moment Consider the beam shown in Fig. 20 . As with the tension-compression bar and the torsion rod. M = EIv ′′ . A moment M is applied at the left end. δκ = δv ′′ and M = M (v ′′ ). L L δκM dx = 0 0 ¯ ¯ δv q dx + M δθ(L) + V δv(L). one has L L δv ′′ EIv ′′ dx = 0 0 ¯ ¯ δv q dx + M δθ(L) + V δv(L) .

we can simply break down the computation of the internal virtual work into the sum of the integrals over sub-parts of the system. 0 From our approximation spaces. 2 . . additively contributes to the total external virtual work. The principle can also be applied far more generally without much added eﬀort. Likewise. if there are multiple types of stress (say bending and torsional). then the virtual work equation states: L ¯ δA [4(x − L) + 2x] EI A [4(x − L) + 2x] dx = M δAL2 0 ∀δA ⇒ A= ¯ M L2 L 0 EI [4(x − L) + 2x]2 dx .The kinematic boundary conditions provide the form for the solution space and the test space as: S = V = v(x) | v(0) = v(L) = v ′ (L) = 0 ′ . In particular. One needs to observe that the internal virtual work is always give by the real stresses times the virtual strains integrated over the volume of the system. since internal virtual work is an integral. If we assume the beam to be linear elastic. δv(x) | δv(0) = δv(L) = δv (L) = 0 Let us consider the following one parameter approximations for S and V ˆ S = ˆ V = v(x) | v(x) = Ax(x − L)2 δv(x) | δv(x) = δAx(x − L) . Both approximation sets are proper subsets of their respective full sets.1 Generalizations So far we have used the Principle of Virtual Work to look at simple problems so that we can better understand what it means and what its properties are. then we need to account for both of them. 21 . we have κ = A [2(x − L) + 2(x − L) + 2x] . ˆ ˆ The set S is parameterized by A and the set V is parameterized by δA. This occurs by simply adding their contributions together. It too is an additive quantity. The same holds for external virtual work. δθ = δA (x − L)2 + 2x(x − L) . Each type of load. if the system has complex shape. δκ = δA [2(x − L) + 2(x − L) + 2x] . 3. For the given problem the relevant virtual work equation is: L ¯ δκM dx = M δθ(0) .

22 . δv2 (x) :Virtual transverse deﬂection of element 2. 2 y 1 L1 x V L2 Figure 13: Angle frame. Write down the appropriate virtual work expression. Consider a cantilevered rod that is subjected to both an end-shear and an end-torque. For the internal virtual work we need to sum the contributions from bending and torsion this gives: L L δv ′′ M dx + 0 0 δϕ′ T dx .Example 6: Bending and torsion. δu2 (x) :Virtual axial displacement of element 2. The virtual work equation is thus L L δv ′′ M dx + 0 0 ¯ ¯ δϕ′ T dx = V δv(L) + T δϕ(L) and the Principle of Virtual Work tells us that for equilibrium this equation must hold for all virtual deﬂections and virtual rotations (in the appropriate test spaces). Let us deﬁne the following internal resultants and virtual ﬁelds: δv1 (y) :Virtual transverse deﬂection of element 1. Here this gives ¯ ¯ V δv(L) + T δϕ(L). Consider the angle frame in Fig. 13. For the external virtual work we need to sum the contributions from each load acting on the system. Element 2 is in a state of axial stress as well as bending and Element 1 is in a state of bending. Example 7: Angle frame.

With these deﬁnitions.M1 (y) :Bending moment in element 1.W. It should be noted that there are no kinematic conditions on δv1 (y) but δu2 (x) and δv2 (x) need to respect the kinematic conditions provided by the built-in support. we can construct the relevant expressions ¯ External V.W. M2 (x) :Bending moment in element 2. = δu′ R2 dx. R2 (x) :Axial force in element 2. = V δv1 (L) L1 L2 ′′ δv1 M1 dy + 0 0 ′′ δv2 M2 dx + 0 L2 Internal V. 2 23 .

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue listening from where you left off, or restart the preview.

scribd